Vertical Drama Shot Length: Why We Aim for 2.5-4 Seconds and Cap Shots at 6
Credits and specs follow TWno1's live billing settings. Last checked: 2026-10-05
Vertical short drama should average 2.5 to 4 seconds per shot, with no single shot over 6, and that is the rule we work to.
No platform publishes that rule. We built it from three inputs of very different strength: a peer-reviewed film study, one industry figure we could not verify, and our own review of a 16-episode demo series with 244 storyboard frames. Here is each one, so you can decide how much weight to give it.
Three numbers, three levels of trust
| Source | Sample | Figure | How far to trust it |
|---|---|---|---|
| Cutting et al., i-Perception (Cornell, 2011) | 160 English-language Hollywood films, 1935-2010 | Average shot length fell from about 10 s in the 1930s-40s to under 4 s after 2000 | Solid, but it measures theatrical film |
| Industry breakdown of top short dramas (single source, not independently verified) | Leading short-drama titles | About 2.5 s average shot, 45.5% of shots under 2 s, about 23 cuts per minute | Direction only |
| Our demo series review (September 2026) | 16 episodes, 244 frames | First edit cut only about once every 5 s | Our own mistake, not a benchmark |
The first row has a gap. We found no published study of average shot length in vertical micro-dramas. Carrying "under 4 seconds" from Hollywood to a phone screen is an assumption, and you should treat it as one.
The second row at least checks out internally. Twenty-three cuts a minute is 60 / 23 ≈ 2.6 seconds per shot, close to the stated 2.5. Internal consistency says nothing about the sample, though. We use it to point the way, not to grade work.
How our first cut got slow
Version one followed a tidy rule. One line of dialogue = one storyboard frame = one camera position = one 5-second clip.
It played badly in two ways. The camera jumped on every line, so a single living room read as three different rooms. And every cut was the same length, which made the episode feel like a slideshow.
Going frame by frame turned up a third problem. Of the 244 frames, 235 had people in them. Only 4 were wide shots or empty shots. And 68% of frames carried no shot-size label at all. When every frame is a similar medium shot of a person, cutting more often does not create rhythm. The viewer sees nearly the same picture each time.
This is where the usual advice falls short. "Just cut faster" assumes length is the only problem. In a 2010 Psychological Science paper on 150 films, Cutting's group found that shot-length sequences in recent films came to approach a 1/f pattern, which the authors link to the rhythm of human attention. Our reading: the issue was uniform length and uniform framing. Shrinking every shot from 5 s to 2.5 s just gives you a faster slideshow.
The fix: plan camera positions first
We flipped the order. For each scene, list 2 to 5 fixed camera positions. Then assign lines to positions, reusing positions freely. Three rules came with it:
- Every cut needs a written reason. No reason, no cut.
- Narration, inner voice and phone audio do not open a new shot. They play over the current one.
- Duration follows content. Short shots run 2.5 s; long ones can run 10.
One scene with 17 lines became a plan of 13 shots built on 8 camera positions. Total runtime was 70.9 s, against at least 87.8 s under the old rule. Unmotivated jump cuts within a camera position: zero. Across all 16 episodes, 244 lines became 175 shots.
That fixed the jumping. It did not fix the pace. 70.9 / 13 ≈ 5.5 s per shot, still well above 4. The next section explains why.
AI clips come out longer than the shot you need
Video models have a minimum clip length, and it is often longer than a line of dialogue. From each vendor's documentation, checked 2026-10-05:
| Model | Official clip length |
|---|---|
| Google Veo 3.1 | 4, 6 or 8 s only |
| Kling 3.0 | 3-15 s |
| Seedance 2.5 (BytePlus) | 4-30 s |
| Vidu Q3 series | 1-16 s (drama and ad variants 3-15 s) |
A four-character Mandarin line takes about a second to say. The clip that carries it comes back at 4 or 5 seconds. Drop it into the edit untouched and the scene looks like slow motion. That is why our average stayed above 5 s.
So we trim, with fixed cut points:
- Dialogue shots: line length + 0.5 s, with ±0.5 s tolerance.
- Shots without dialogue: 2-3 s.
- Never cut earlier than 0.5 s after the line ends.
The last rule came from a mistake. One trimmed shot lost the final syllable of its line. Another opened on 1.9 seconds of silence. On the same set of 7 shots, the old way met the length target 0 times; with the three cut points above, all 7 passed. If a line will not fit in 6 seconds, shorten the line or let the shot run long. Do not chop the audio.
A pacing checklist you can run this week
| Check | Threshold | If it fails |
|---|---|---|
| Average shot length | 2.5-4 s | Trim dead tails after dialogue first, then consider splitting |
| Any single shot | ≤ 6 s | Mark deliberate long takes; otherwise cut the line or split |
| First shot of each scene | Extreme wide or full shot | Add an establishing shot |
| Neighboring shots | Different shot sizes | Switch to over-the-shoulder, reaction or close-up |
| Shots with no person, or a very small one | ≥ 25% of the episode | Add empty shots, props, hand details |
| Frames with no shot-size label | 0 | Label them in the storyboard |
| Opening | Hook inside 6 s | Move the conflict earlier |
The 6-second hook comes from TikTok's Creative Codes for advertisers, which says 90% of ad-recall impact and 80% of awareness impact land within the first 6 seconds. That is ad data, not drama data. The direction still holds. For episode length we plan 60-120 s; a China Economic Weekly report carried by People's Daily Online (2024-12-15) describes today's micro-dramas as vertical series of about 1-2 minutes per episode.
When these thresholds are not a good fit
- Horizontal long-form. These numbers are for phone-first vertical drama. Even theatrical film only dropped below a 4 s average after 2000.
- Long dialogue. A 30-character Mandarin line needs about 7.5 s at roughly 4 characters per second. Holding a 6 s cap would clip it. Rewrite the script instead of forcing the edit.
- Intentional long takes. A confession or a silence needs room. Mark it in the shot list and move on.
- The 2.5 s figure has one source. If your distributor has its own data, use theirs.
Look back at the 68% of frames with no shot size and the wall of medium shots. You could see that on the storyboard, long before any video existed. In TWno1, every shot's storyboard image waits for your approval before video generation starts, so fixing two same-size neighbors costs you one image, not one clip. Generation runs 4-30 s per clip; shots download as ordinary files, so you can trim to "line plus 0.5 s" in your own editor. To lay out one scene against the checklist above, open a project in Studio.
Related reading: Face-Similarity Scores for AI Drama Characters: Four Checks and the Thresholds We Use; Keeping One Character Consistent Across a Whole AI Series: References, Prompts and a Per-Shot Look Sheet; What Is an AI Short Drama? Definition, Workflow, and Real Error Rates.
FAQ
What is a good average shot length for vertical short drama?
We use a 2.5-4 second average with a 6-second cap per shot. The 4-second end comes from a Cornell study of 160 Hollywood films, where the average fell below 4 s after 2000. The 2.5-second end is an industry breakdown of top short dramas from a single, unverified source.
My AI video clip is 5 seconds but the line is short. What should I do?
Trim it. Cut dialogue shots to line length plus 0.5 seconds and shots without dialogue to 2-3 seconds, and never cut sooner than 0.5 seconds after the line ends. Google's Veo 3.1 only offers 4, 6 or 8 second clips, so trimming in post is normal.
What is a camera-position plan in short drama?
A camera-position plan fixes 2 to 5 camera positions per scene before any shot is drawn, then assigns each line of dialogue to one of them. In our demo series, a 17-line scene went from at least 87.8 seconds to 70.9 seconds and 13 shots, with zero unmotivated jump cuts.
Is faster cutting always better for short drama?
No. Cutting's 2010 study of 150 films found shot lengths in recent films vary in a 1/f-like pattern rather than staying uniform. In our 244-frame review, uniform 5-second shots and near-identical medium framing were both problems; shortening every shot without changing framing produces a faster slideshow.
How long should one vertical drama episode be?
We plan 60-120 seconds per episode. A 2024-12-15 China Economic Weekly report on People's Daily Online says micro-dramas usually run about 1-2 minutes per episode. At a 2.5-4 second average, a 90-second episode holds roughly 23 to 36 shots.
References
- Cutting et al. (2011), i-Perception: shot length in 160 Hollywood films, 1935-2010 (checked 2026-10-05)
- Cutting, DeLong & Nothelfer (2010), Psychological Science: shot-length sequences in 150 films (checked 2026-10-05)
- TikTok Creative Codes (PDF) (checked 2026-10-05)
- Google Gemini API: Veo video generation (checked 2026-10-05)
- Kling 3.0 user guide (checked 2026-10-05)
- BytePlus ModelArk: Seedance 2.5 (checked 2026-10-05)
- Vidu model map (checked 2026-10-05)
- People's Daily Online / China Economic Weekly, 2024-12-15 micro-drama report (checked 2026-10-05)